基于模糊信息熵的彩铃客户属性约简算法

针对彩铃业务交易记录海量化和客户属性高维化、混合性的特点,提出一种应用模糊信息熵对彩铃客户属性进行约简的方法。以两个地区的营销返回样本为原始数据,结合CART分类器和10折交叉验证,比较由文中方法和经典粗糙集属性约简方法得到的特征数量和分类精度。实验结果表明,文中方法获得了相对较高的平均分类精度(81.36%)和最少的平均特征属性(4.5个),有效地约简了彩铃客户属性并改善了分类能力。

Attribute reduction algorithm of color ring customers based on fuzzy information entropy

Aimed at the immense dealing track record of the color ring operation,and the high-dimensional and hybrid properties of the customers attribute,this paper proposes an attribute reduction method of the color ring customers based on fuzzy information entropy.The paper describes the combination of the CART classifier with 10-fold cross validation and the comparison between the number and accuracy of selected features using our method and classical rough,based on the marketing data sets of two different cities.Results show that the proposed method gives a much better average classification accuracy（81.36%）,and the least average feature attribute（4.5）,thus effectively reducing attribute,and even improving classification power.